Detection of uterine fibroids using wavelet packet features with BPNN classifier

Uterine fibroids also referred as leiomymas are the most common tumors persist within the wall of the female genital tract. This abnormality is predominant among woman of childbearing age where the secretion of estrogen hormone is significant. The most crucial factor is that the presence of fibroid can cause infertility and repeated miscarriage. In the recent years, ultrasonic imaging found to be an appropriate tool for diagnosis of uterus related disorders. This paper presents an automated detection of uterine fibroid by using wavelet features and a neural network classifier. Based on user-defined ROI, a three level wavelet packet decomposition is applied to calculate the vertical and horizontal coefficients. In order to distinguish the normal and fibroid uterus images, a feed forward backpropogation neural network(BPNN) classifier is used and the performance are evaluated in terms of sensitivity, specificity and classification accuracy. It is observed from the experimental study that a classification accuracy of 95.1% is achieved which indicates the suitability of the proposed scheme for clinical evaluation